[E-voting] Not random enough?

Michael McMahon michael at hexmedia.com
Mon Mar 15 16:03:18 GMT 2004

I think there is a more fundamental problem. The whole approach
taken by PTB to testing this system is *unscientific*.
(I think Adrian agrees with me on this)

For an organisation that claims to be "the highest technical authority
in the Federal Republic of Germany for the field of Metrology..."
this is quite a serious accusation to fling.

The approach they took was to only test what they could observe visually
on the machines and to rely on inspection of the source code for 
everything else.

So, for instance, probably the most fundamental requirement of the 
voting machine
is #(45) in section 8.1 "A vote recorded in the primary ballot module must
be the vote that the voter has cast" is only tested by looking at the 
source code.
This is not a test at all. This requirement should have been tested by 
the ballot module to some independent test-tool and checking if data 
entered on
the voting machine is correctly stored on the module.

Someone made the analogy last week, of testing a voltmeter. So, instead of
testing the voltmeter by using an externally verifiably voltage, they 
would have
tested it, by taking the voltmeter apart and looking at its components.

Therefore, what PTB did, could not in any way be described as a functional
test of the voting machines.

- Michael

Aengus Lawlor wrote:

>Adrian Colley <aecolley at spamcop.net> wrote:
>>Subject: [E-voting] > To: Colm MacCarthaigh <colm at stdlib.net>
>>A new thing is that I looked at the PTB's evaluation of the random
>>placement of ballot records within the ballot module.  It _isn't_
>>random enough.  If you know the preferences of voters #1 and #3, and
>>they're unique, then you can deduce voter #2's preferences with 100%
>>accuracy just from examining the order of ballot records on the
>>ballot module.  It doesn't matter how many ballots are stored on the
>>ballot module.  Apparently 80% of all ballots are unique.  I think
>>this is a real honest-to-God showstopper for the ESI2.
>Can you expand on that a bit, Adrian?
>E-voting mailing list
>E-voting at lists.stdlib.net

More information about the E-voting mailing list